Mills@udel.edu (11/06/90)
Philip, See my paper in the January 1989 ACM Computer Communications Review for a discussion of NTP timekeeping accuracy. If that tart rag is not at convenient paw distance, a PostScript copy is in the file pub/ntp/ccr.ps on louie.udel.edu. From data collected here, a systematic offset of a few milliseconds is not unlikely, given the network connectivity under the Atlantic. Dave
philip@beeblebrox.dle.dg.com (Philip Gladstone) (11/06/90)
I have been running xntp for some time now with a couple of local
stratum 1 clocks and some remote clocks. I have noticed that there
seems to be a persistent difference of a few millisecs between my
clocks and some of the US clocks. [I'm in the UK!].
I happen to know that my clocks are not very good at finding the
actual 'start-of-second' pulse -- analog circuitry and all that.
However, I beleive that the edge that I get is within 1 ms of real
time.
The question is:
How correct are the stratum 1 clocks? Is the precision an accurate
estimate?
Does anyone have any neat tools for recording time difference
information over a period of a few weeks to construct graphs etc?
--
Philip Gladstone
Development Lab Europe
Data General, Cambridge
England. +44 223-67600